Table 5.

Multivariate analysis of biomarkers and their interactions to predict radiographic and erosive progression

GLM	ΔSvH estimate (SE)	ΔErosion estimate (SE)
14-3-3η	0.098 (0.045) *	0.088 (0.033) **
Age (years)	0.106 (0.017) ***	0.056 (0.013) ***
Anti-CCP2	0.003 (0.002)	0.001 (0.002)
RF	0.000 (0.001)	0.000 (0.001)
Anti-Sa positive	3.104 (0.747) ***	2.878 (0.543) ***
CRP	0.032 (0.008) ***	0.025 (0.006) ***
GEE	ΔSvH ≥5 RR (95 % CI)	ΔErosion ≥5 RR (95 % CI)
14-3-3η ≥0.50 ng/ml	2.176 (1.193–3.970) *	2.276 (1.448–3.577) ***
Age ≥65 years	2.680 (1.859–3.864) ***	2.751 (1.760–4.301) ***
Anti-CCP2-positive	1.048 (0.696–1.577)	1.134 (0.635–2.025)
RF-positive	1.191 (0.739–1.921)	1.689 (1.042–2.738) *
Anti-Sa-positive	1.537 (0.708–3.335)	2.507 (1.348–4.659) **
CRP >8.0 mg/L	1.425 (1.112–1.826) **	1.834 (1.304–2.578) ***
14-3-3η ≥0.50 * age ≥65 years	0.677 (0.396–1.156)	Not included
14-3-3η ≥0.50 * RF-positive	0.724 (0.386–1.358)	0.552 (0.310–0.983) *
14-3-3η ≥0.50 * anti-Sa positive	0.898 (0.470–1.717)	0.400 (0.204–0.786) **
Age ≥65 * anti-CCP2 positive	0.899 (0.507–1.596)	0.692 (0.351–1.365)
Age ≥65 * anti-Sa positive	0.593 (0.345–1.017)	Not included
RF-positive * anti-Sa positive	1.256 (0.607–2.597)	Not included
Anti-CCP2-positive * anti-Sa positive	Not included	0.856 (0.424–1.728)

Age, gender, 14-3-3η, C-reactive protein (CRP) and rheumatoid arthritis-associated rheumatoid factor (RF), antibodies against citrullinated peptides, second generation (anti-CCP2) and anti-Sa antibodies, and all their interaction terms were included in multivariate generalized linear model (GLM) or general estimating equation (GEE) analysis. Only the models with the lowest Akaike information criterion (AIC) or quasi-AIC are presented. *p <0.05, **p <0.01, ***p <0.001. SE standard error ΔSvH progression in the total Sharp score modified by van der Heijde, ΔErosion progression in the erosion component of the Sharp score modified by van der Heijde. Not includedvariable excluded from the final model